The standard for intervening in a refereeing decision is usually phrased in plain language rather than in numbers. That phrasing is doing precise work despite sounding loose.
The wording describes strength of evidence
A clear error is one that replay resolves decisively, where a reasonable observer watching the footage would reach the same conclusion. The emphasis sits on what the footage shows rather than on how much the outcome mattered.
The test is not whether the reviewer disagrees with the referee. It is whether the footage removes the disagreement.
Framed that way, the phrase is a description of evidence quality rather than a description of how severe the incident was.
It draws a line between two kinds of error
Some mistakes come from missing something entirely, such as an off-the-ball incident behind play. Replay supplies information the referee never had, and closing that gap is what video review was built for.
Others come from weighing something differently, such as judging the force of a challenge. Replay supplies the same information again.
The threshold is designed to catch the first kind and leave the second with the on-field official.
Consistency depends on training, not the words
Any verbal standard needs a shared interpretation before it produces consistent outcomes across officials and competitions. Words do not carry their own calibration, and two officials can apply the same sentence differently for years without noticing.
That shared reading is built through calibration sessions where officials review the same incidents and compare where they would intervene.
Without that process the same wording produces very different intervention rates, which is the usual explanation for cross-league inconsistency.
Camera angles change what counts as clear
Evidence quality depends on coverage. A tackle captured from one distant angle may be genuinely ambiguous on replay, and no amount of slowing the footage recovers detail the camera never captured.
The same incident with a close side-on view can be unambiguous, so the identical challenge may or may not meet the threshold depending on the production.
This means the standard is partly a function of broadcast resources, which vary between competitions and between matches within them.
Numerical thresholds would not fix it
Replacing the phrase with a numeric confidence level would look more rigorous but require a measurement that does not exist for subjective calls.
Factual questions such as offside already use measured thresholds, and those are the decisions where automation has advanced furthest. Position and time have units; recklessness does not.
For everything else the standard has to remain a description of evidence, applied by trained officials, because the underlying judgement has no unit.